Problem: Create 'uint8' type tensor from float data raise RuntimeError;
Fix:  Use same internal data type for 'uint8' and 'bool' tensor and allows
convert float data to bool tensor.